For years the safe answer was buy unless it is your core differentiator. AI has lowered the barrier to building, while the hidden cost of off-the-shelf software — the 50 to 80 percent fit problem — never went away. Here is how to decide now.
Every organization adopting AI faces the same fork at every capability: build it custom, buy it off the shelf, or do not do it at all. The framework for answering that question has not changed. What has changed is where the line falls — and most teams are still using the old map.
The old rule, and why it existed
For a long time the safe default was: buy unless the capability is your core differentiator. That rule made sense because building was slow and expensive. Standing up custom software meant a long engineering cycle, a team to maintain it forever, and real risk it never shipped. Against that, a monthly subscription looked like the obvious call for anything that was not central to your edge.
That rule was right for its moment. The moment has moved.
Why the line is shifting
The barrier to entry to building has dropped sharply. With the right people and the right partner, custom software that used to take a quarter and a team can now take a fraction of that. This is not about it being cheap — it is about it being feasible. Things that were never worth building because the effort dwarfed the payoff are now squarely on the table.
The result is simple: more capabilities are worth building than the old rule suggests. Not everything. But the bar that a custom build has to clear is lower than it was, because the cost of clearing it is lower.
The hidden cost of buying: the 50 to 80 percent problem
Here is what the old “just buy it” instinct quietly ignores. Off-the-shelf software almost never fits how you actually work. In practice a tool matches your real needs somewhere between 50 and 80 percent — and then you pay for the gap one of two ways.
Either you bend your processes to the tool, changing how your team works to match what the software assumes. Or you customize the tool to match your processes, with configuration, integration, workarounds, and the consultants who do that work. It happens almost every time, because there is no perfect one-size-fits-all product. A vendor builds for the average of a thousand customers; you are not the average.
That gap is a real cost. It is just easy to miss because it does not show up on the invoice — it shows up in adoption friction, in the workflow that never quite fits, and in the customization bill you pay anyway. Once you count it honestly, the math on buying looks different than the sticker price suggests.
So what should you still buy?
Plenty. Genuinely commodity capabilities — transcription, generic chat, standard OCR, common integrations — are solved well by vendors who have spent years on them, and where an off-the-shelf tool fits your needs cleanly, buy it and move on. The point is not “build everything.” It is that the set of things worth building has grown, and the 50 to 80 percent fit gap should be priced into every buy decision instead of waved away.
And building is still something you own. A custom system is a system you maintain as models, data, and your business change underneath it. That is exactly why the right partner matters — not just to build the thing, but to keep it honest and current after it ships. Building is more accessible than ever; it is not free of responsibility.
The questions to ask per capability
For each candidate capability, ask:
- How well does the best off-the-shelf tool actually fit? Be honest about the gap, and price in what closing it will cost in process change or customization.
- Is this close to how we win? The nearer a capability is to your real edge, the more the fit gap hurts and the more a tailored build pays off.
- Do we have the partner to build and maintain it? Lower barrier to entry is only real if someone keeps the system alive after launch.
Run those three across your roadmap and the answer usually sorts itself: build the things where a tailored fit genuinely matters and you have the partner to sustain them, buy the clean-fit commodities, and skip the rest. That discipline is still the fastest way to avoid tool sprawl and wasted budget — the line between build and buy has just moved toward build.
If you want a build-buy-skip recommendation for each capability on your roadmap, that is what our Build vs Buy advisory delivers.